As increasing numbers of people turn to generative AI chatbots for therapy and companionship (HBR recently listed this as genAI’s #1 use case), we’re long overdue to examine this question beyond the surface level.

Instead of asking ourselves whether therapy bots can approximate elements of social interaction and psychological care, why not progress to questions of whether decontextualized elements of therapy are enough, or whether humans should rely on machines to solve our warm-blooded, socially-informed struggles?

Supportiv CEO Helena Plater-Zyberk explores these questions and the evidence that can lead us to an answer – that “thera-bots” are likely not the best solution for population-scale mental health needs. Recent research supports this conclusion in practice, such as a study through OpenAI and MIT Media Lab which reveals frequent chatbot use increases loneliness and reduces likelihood of pursuing real-life social interaction. Plater-Zyberk’s piece in MedCity News delves into potential reasons why:

  • Most of our society’s mental health struggles have social or attachment-related roots
  • Positive human-to-human experiences are a prerequisite for healing social and attachment-related wounds
  • The types of therapy chatbots show promise in providing is not the gold standard for attachment-related struggles
  • Chatbots pose a real risk of artificial intimacy, associated with reduced self esteem other negative outcomes (see aforementioned OpenAI and MIT Media Lab study)
  • Chatbots offer comfort over healing

Additionally, Plater-Zyberk goes on to attest that many of the needs behind bot use – low cost, immediate 24/7 access, promises of nonjudgment – can also be met by human-first solutions like live-moderated digital peer support. Supportiv operates on this truth, with outcome metrics to prove it.

Learn more here.